Image-Based Field Boundary Detection and Identification

1. A method of detecting and identifying a boundary of a field, comprising: defining positional coordinates of a seed point in a particular field, and a first bounding box around the seed point; obtaining a plurality of images over time that include the first bounding box around the seed point; clipping the plurality of images so that pixels in each image in the plurality of images refer to positional coordinates of the first bounding box; building a mask for a boundary of the particular field by identifying one or more edges of the particular field from the plurality of images; filling an area around the seed point and within the mask that has pixels of a similar color to pixels representing the seed point in each image in the plurality of images; checking the filled area around the seed point by a) defining a second boundary box for the filled area, b) determining whether threshold parameter values representing upper and lower distances from a pixel brightness of the seed point are satisfied in the second boundary box for the filled area, and c) tuning the threshold parameters values where the pixel brightness is outside the threshold parameter values to adjust the filled area; combining, for each image, the filled area into a combined flooded set; computing a boundary around the combined flooded set; and translating the boundary in the combined flooded set to positional coordinates in the particular field.

2. The method of claim 1 , wherein the plurality of images are obtained from at least one of satellites, field-based robots, aircraft, and remotely-piloted vehicles over time.

3. The method of claim 1 , wherein the plurality of images represent the same location captured at different times, over a period of time spanning either a current season or over multiple years.

4. The method of claim 3 , further comprising creating an average image from the plurality of images by computing average pixel color, and smoothing each image to eliminate small-scale noise.

5. The method of claim 1 , further comprising matching the seed point with the plurality of images.

6. The method of claim 1 , further comprising interactively tuning the boundary in the combined flooded set by selecting a wrongly-filled area in the plurality of images, and determining whether the wrongly-filled area is inside the boundary.

7. The method of claim 6 , further comprising creating an augmented mask with one or more points known to be inside the actual boundary where the wrongly-filled area is inside the boundary and the wrongly-filled area is maintained as outside an actual boundary, re-filling the area around the seed point using the augmented mask, and subtracting a re-filled combined flooded set from the existing combined flooded set, and calculating a revised boundary of the particular field.

8. The method of claim 6 , further comprising creating an augmented mask with one or more points known to be outside the actual boundary, and adding the bounding box, where the wrongly-filled area is outside the boundary and the wrongly-filled area is maintained as inside an actual boundary, and the method, re-filling the area around the seed point using the augmented mask, and adding a re-filled combined flooded set to the existing combined flooded set, and calculating a revised boundary of the particular field.

9. The method of claim 6 , further comprising straightening the boundary by one or more of selecting two different boundary points, and selecting a position near the boundary.

10. A method, comprising: identifying 1) a seed point representing a particular field, and 2) a bounding box around the seed point, the bounding box having a plurality of threshold parameters defining a pixel brightness relative to pixels representing the seed point, and a plurality of images over time of the particular field that include bounding box and the seed point; analyzing the input data in a plurality of data processing modules within a computing environment in which the plurality of data processing modules are executed in conjunction with at least one processor, the data processing modules configured to detect and identify a boundary of the particular field, by: modifying the plurality of images so that pixels in each image in the plurality of images refer to positional coordinates of the bounding box, detecting one or more edges of the particular to create a mask for a boundary of the particular field, performing a floodfill operation to flood an area around the seed point within the mask, that has pixels of a similar color to pixels representing the seed point in each image in the plurality of images, comparing pixel brightness of the flooded area with the plurality of threshold parameters, and tuning values in the plurality of threshold parameters representing upper and lower distances from a pixel brightness of the seed point where the pixel brightness exceeds the values to adjust the size of the flooded area, creating a combined flooded set from the flooded areas in each image, applying an alpha-shape operation to build a boundary around the combined flooded fill set; and generating, as output data, a set of positional coordinates representing the boundary of the particular field.

11. The method of claim 10 , wherein the plurality of images are obtained from at least one of satellites, field-based robots, aircraft, and remotely-piloted vehicles over time.

12. The method of claim 10 , wherein the plurality of images represent the same location captured at different times, over a period of time spanning either a current season or over multiple years.

13. The method of claim 10 , further comprising creating an average image from the plurality of images by computing average pixel color, and applying a median filter to eliminate small-scale noise in each image.

14. The method of claim 10 , further comprising matching the at least one seed point with the plurality of images.

15. The method of claim 10 , further comprising interactively tuning the boundary in the combined flooded set by selecting a wrongly-filled area in the plurality of images, and determining whether the wrongly-filled area is inside the boundary.

16. The method of claim 15 , further comprising creating an augmented mask with one or more points known to be inside the actual boundary where the wrongly-filled area is inside the boundary and the wrongly-filled area is maintained as outside an actual boundary, re-filling the area around the seed point using the augmented mask, and subtracting a re-filled combined flooded set from the existing combined flooded set, and calculating a revised boundary of the particular field.

17. The method of claim 15 , further comprising creating an augmented mask with one or more points known to be outside the actual boundary, and adding the bounding box, where the wrongly-filled area is outside the boundary and the wrongly-filled area is maintained as inside an actual boundary, and the method, re-filling the area around the seed point using the augmented mask, and adding a re-filled combined flooded set to the existing combined flooded set, and calculating a revised boundary of the particular field.

18. The method of claim 15 , further comprising straightening the boundary by one or more of selecting two different boundary points, and by electing a position near the boundary.

19. A system, comprising: a computing environment including at least one computer-readable storage medium having program instructions stored therein and a computer processor operable to execute the program instructions to perform an image-based field boundary detection in a particular field within a plurality of data processing modules, the plurality of data processing modules including: a data identification component configured to determine 1) a seed point representing a particular field, and 2) a bounding box around the seed point, the bounding box having a plurality of threshold parameters defining a pixel brightness relative to pixels representing the seed point; an image retrieval component, configured to obtain a plurality of images over time of the particular field that include bounding box and the seed point; a plurality of data processing modules configured to detect and identify a boundary of the particular field based on the seed point from the plurality of images, the plurality of data processing modules configured to: modify the plurality of images so that pixels in each image in the one or more images refer to positional coordinates of the bounding box, detect one or more edges of the particular to create mask for a boundary of the particular field, perform a floodfill operation to flood an area around the seed point that has pixels of a similar color to pixels representing the seed point in each image in the plurality of images, compare pixel brightness of the flooded area around the seed point with the plurality of threshold parameters, and tuning values in the plurality of threshold parameters representing upper and lower distances from a pixel brightness of the seed point where the pixel brightness exceeds the values to adjust a size of the flooded area, create a combined flooded set from the flooded areas in each image, apply an alpha-shape operation to build a boundary around the combined flooded fill set; and a translation module configured to generate a set of positional coordinates representing the boundary of the particular field.

20. The system of claim 19 , wherein the plurality of images are obtained from at least one of satellites, field-based robots, aircraft, and remotely-piloted vehicles over time.

21. The system of claim 19 , wherein the plurality of images represent the same location captured at different times, over a period of time spanning either a current season or over multiple years.

22. The system of claim 19 , wherein the data processing modules are further configured to create an average image from the plurality of images by computing average pixel color, and apply a median filter to eliminate small-scale noise in each image.

23. The system of claim 19 , wherein the data processing modules are further configured to match the at least one seed point with the plurality of images.

24. The system of claim 19 , wherein the data processing modules are further configured to interactively tune the boundary in the combined flooded set by selecting a wrongly-filled area in the plurality of images, and determining whether the wrongly-filled area is inside the boundary.

25. The system of claim 24 , wherein the data processing modules are further configured to create an augmented mask with one or more points known to be inside the actual boundary where the wrongly-filled area is inside the boundary and the wrongly-filled area is maintained as outside an actual boundary, re-filling the area around the seed point using the augmented mask, and subtracting a re-filled combined flooded set from the existing combined flooded set, and calculating a revised boundary of the particular field.

26. The system of claim 24 , wherein the data processing modules are further configured to create an augmented mask with one or more points known to be outside the actual boundary, and adding the bounding box, where the wrongly-filled area is outside the boundary and the wrongly-filled area is maintained as inside an actual boundary, and the method, re-filling the area around the seed point using the augmented mask, and adding a re-filled combined flooded set to the existing combined flooded set, and calculating a revised boundary of the particular field.

27. The system of claim 24 , wherein the data processing modules are further configured to straighten the boundary by one or more of selecting two different boundary points, and by electing a position near the boundary.

28. The system of claim 27 , wherein the data processing modules are further configured to further deciding which boundary points are used for straightening the boundary based on user interaction with the plurality of images.